Elsevier

Journal of Hydrology

Volume 512, 6 May 2014, Pages 107-125
Journal of Hydrology

Evaluation of multi-model simulated soil moisture in NLDAS-2

https://doi.org/10.1016/j.jhydrol.2014.02.027Get rights and content

Highlights

  • First time evaluation of long-term soil moisture products in NLDAS-2 using three in situ observations.

  • Evaluation was performed for multiple models, soil layers and metrics.

  • Provide solid analysis and support for use of soil moisture in U.S. drought monitor.

Summary

The North American Land Data Assimilation System (NLDAS) phase 2 (NLDAS-2) has generated 31-years (1979–2008) of water and energy products from four state-of-the-art land surface models (Noah, Mosaic, SAC, VIC). The soil moisture data from these models have been used for operational drought monitoring activities, but so far have not yet been comprehensively evaluated. In this study, three available in situ soil moisture observation data sets in the United States were used to evaluate the model-simulated soil moisture for different time scales varying from daily to annual. First, we used the observed multiple layer monthly and annual mean soil moisture from the Illinois Climate Network to evaluate 20-years (January 1985–December 2004) of model-simulated soil moisture in terms of skill and analysis of error statistics. Second, we utilized 6-years (1 January 1997–31 December 2002) of daily soil moisture observed from 72 sites over the Oklahoma Mesonet network to assess daily and monthly simulation skill and errors for 3 model soil layers (0–10 cm, 10–40 cm, 40–100 cm). Third, we extended the daily assessment to sites over the continental United States using 8-years (1 January 2002–31 December 2009) of observations for 121 sites from the Soil Climate Analysis Network (SCAN). Overall, all models are able to capture wet and dry events and show high skill (in most cases, anomaly correlation is larger than 0.7), but display large biases when compared to in situ observations. These errors may come from model errors (i.e., model structure error, model parameter error), forcing data errors, and in situ soil moisture measurement errors. For example, all models simulate less soil moisture due to lack of modeled irrigation and ground water processes in Illinois, Oklahoma, and the other Midwest states.

Introduction

Soil moisture is an important variable in the climate system in part because it has a long memory of climate signals over the continents. It also plays a critical role in shaping ecosystem response to the physical environment and controls the partitioning of available energy at land surface into sensible and latent heat exchanges with the overlying atmosphere, thus modulating the water and energy balances through the soil moisture and temperature states. The important influence of soil moisture on weather and climate has been demonstrated by Reed, 1925, Namias, 1952, Delworth and Manabe, 1988, Delworth and Manabe, 1993. In recent years, sensitivity studies of climate models have shown soil moisture contribution to precipitation predictability (Dirmeyer, 2000), in particular in transition zones between dry and humid climates (Koster et al., 2003, Koster et al., 2004). Therefore, improving weather and seasonal climate prediction requires reliable soil moisture as the initial state of climate models.

Since in situ soil moisture observations are limited in both time and space, model-simulated soil moisture products often serve as alternatives (i.e., for providing initial conditions and/or inputs) for weather and climate models, crop models, ecosystem models, agricultural drought monitoring, and flood monitoring (Robock et al., 2000). Soil moisture products retrieved from satellite-based remote sensing, such as AMSR-E (Advanced Microwave Scanning Radiometer for the Earth Observing System; Reichle et al., 2007), ASCAT (Advanced Scatterometer; Wagner et al., 1999) and SMOS (Soil Moisture and Ocean Salinity; Kerr et al., 2011), offer potential, however, they are generally useable only in the upper few centimeters over sparsely vegetated area for microwave sensors, although efforts are being made to extrapolate these measurements to the root zone (Gao et al., 2007, Reichle et al., 2007, Sabater et al., 2007, Drusch et al., 2009). Thermal-based retrievals also offer potential to infer root-zone soil moisture (Hain et al., 2009) and can complement microwave retrievals (Li et al., 2010). Model-based soil moisture products can be classified as (Robock et al., 2000): (1) climate models run with climatological boundary, (2) climate models run with observed sea surface temperatures, (3) weather forecast models run in reanalysis mode, (4) weather forecast models run in real-time forecast mode, and (5) stand-alone land surface models run with specified forcing at specific locations. Real-time forecast systems may also be differentiated by what they assimilate, such as screen-level data in the ECMWF (European Center for Medium-Range Weather Forecasts) operational system (Drusch and Viterbo, 2007) or satellite soil moisture retrievals in the UKMO (United Kingdom Met Office) operational global model (Dharssi et al., 2011). A stand-alone model run is also called an offline or uncoupled model run because the atmospheric forcing data are predetermined without land–atmosphere interactions. The offline run has the advantage that it can be used as a testing tool because it is easier to compare models when all the models are forced with same and more realistic forcing.

The North American Land Data Assimilation System phase 1 (NLDAS-1, Mitchell et al., 2004), with support from GEWEX (Global Energy Water and Energy Experiment) Continental-Scale International Project (GCIP), established the NLDAS configuration and framework. The goal of the NLDAS was to provide realistic land surface initial conditions (including soil moisture) for improving numerical weather prediction. Since its initiation over a decade ago, the goals and applications of the NLDAS have diversified into evaluating land surface models, providing a test-bed for assimilation of satellite remote sensing data, drought monitoring and seasonal hydrological prediction, among others. The NLDAS is aligned with type 5 soil moisture products described above, in that it runs stand-alone land surface models with specified forcing over the NLDAS domain, which includes the continental United States, southern Canada, and northern Mexico.

NLDAS-1 selected four land surface models (Noah, Mosaic, SAC, and VIC). These models represent different approaches to land surface modeling. The Noah model is the land model of the NCEP (National Centers for Environmental Modeling Prediction) operational regional and global weather and climate models (Chen et al., 1997, Betts et al., 1997, Ek et al., 2003). The Mosaic model is the land model for the NASA global climate model (Koster and Suarez, 1994, Koster and Suarez, 1996), but has been replaced by the Catchment land surface model for the recent upgrade of NASA’s GOES-5 system. The VIC model was developed as a macroscale semi-distributed model (Liang et al., 1994, Wood et al., 1997). The SAC model was developed as a semi-distributed hydrological model (Koren et al., 1999) based on a lumped conceptual hydrology model (Burnash et al., 1973), calibrated for small catchments and used operationally in National Weather (NWS) Service River Forecast Centers (RFCs). The first two models were developed within the surface–vegetation–atmosphere transfer scheme community for coupled atmospheric modeling with focus on the interaction between land and atmosphere through surface energy and water fluxes exchanging processes. The last two models were developed by the hydrological community as uncoupled hydrological models with focus on hydrologic prediction such as for streamflow.

The NLDAS-1 ran the four land surface models from 1 October 1996 to 30 September 1999 using gauge observed precipitation and other forcing data derived from the Eta model to generate output water fluxes, energy fluxes, and state variables. The simulated soil moisture for October 1997 to September 1999 was evaluated using in situ observed soil moisture in Illinois (Schaake et al., 2004) and Oklahoma Mesonet network (Robock et al., 2003), as well as for other hydrological variables as summarized by Mitchell et al. (2004). For Illinois, the Noah and SAC models agreed with observations in both storage range and magnitude, Mosaic yielded a greater storage range, and VIC had lower storage magnitude than observations (Schaake et al., 2004). For Oklahoma, comparison between simulated and observed 40 cm soil moisture averaged over all 72 Mesonet stations presented substantial differences in magnitude among the four land surface models and between model-simulated results and observations, with VIC showing the best agreement with observations. However, there was good agreement between simulated and observed soil moisture anomalies when the mean annual soil moisture values were removed for the observation and each individual model (Robock et al., 2003).

Although the NLDAS-1 analyses provided useful information on model performance, the two-year period of the simulations is too short to provide reliable information on model performance across a range of climate variability. It is also too short to compute robust anomalies and percentiles that are useful for drought analysis and drought monitoring, such as used to monitor agricultural drought in the National Integrated Information System (NIDIS), which is one of main motivations to develop the NLDAS project. With the support from NOAA’s Climate Prediction Program of the Americas (CPPA), the NLDAS-1 has been recently extended to NLDAS-2 using upgraded land surface models, more accurate meteorological forcing (Xia et al., 2012a), and a set of 30-year retrospective simulations covering from 1 January 1979 to 31 December 2008. NLDAS-2 has been running in realtime from 1 January 2009 to present with a 4-day lag. The long-term soil moisture products from these simulations, which have high temporal (hourly) and spatial (1/8th degree) resolution, have so far not been evaluated using available in situ soil moisture observations. The purpose of this paper is to evaluate the simulated soil moisture from the four NLDAS-2 land surface models against a set of in situ soil moisture measurements. Among possible sources of data, the Illinois soil moisture database, Oklahoma Mesonet soil moisture data set, and SCAN soil moisture data set provide reasonably long-term and reliable data. The Illinois network provides a long-term monthly mean soil moisture dataset, the Oklahoma Mesonet provides a recent 6-year (1997–2002) soil moisture dataset with a high temporal and spatial resolution, and the SCAN has a 8-year (2002–2009) daily soil moisture observation covering multiple locations across the continental United States. These observations allow us to evaluate and validate the simulated soil moisture from NLDAS-2. In this paper we first describe the three observed soil moisture data sets and the NLDAS model-simulated soil moisture data sets. Next we present comparisons between simulated and observed soil moisture at different depths and time scales, and then we give a summary and some discussion in the last section.

Section snippets

Observations

Three different sets of in situ soil moisture observations were used to analyze the simulation skill (anomaly correlation) and error of the modeled soil moisture. The first set is from the Illinois network of soil moisture observations (Hollinger and Isard, 1994), which consists of 18 sites throughout the state of Illinois from January 1981 to June 2005. Because the evolution of the observed soil moisture from 1981 to 1984 is clearly different from the later observations (Meng et al., 2012) and

Comparisons of simulated and observed soil moisture

Comparisons of simulated and observed soil moisture can be done in two ways. The first is a direct comparison between the simulated soil moisture and observations at each individual station, but this kind of comparison may suffer from the differences in scale (Crow et al., 2011). Soil moisture spatial variation is related to small-scale hydrological processes, soil characteristics and vegetation cover and is very heterogeneous (Vinnikov et al., 1996, Crow and Wood, 1999, Entin et al., 2000).

Summary and discussion

The validation against the soil moisture observations over the Illinois and Oklahoma Mesonet networks shows that all four land surface models can capture the broad features of observed soil moisture variations, such as the seasonal cycle and interannual variability, although the temporal variability of the simulated soil moisture is higher than observed. The models are also able to well simulate the daily, monthly and annual soil moisture anomalies and capture most of the wet and dry events at

Acknowledgements

This work by NCEP/EMC was supported by the Climate Program Project of the Americas (CPPA) of NOAA Climate Program Office (CPO) as the core project for the EMC (Y. Xia, H. Wei, J. Meng) and the Office of Hydrological Development (J. Dong). We thank the NOAA Office of Global Program and NASA Land Surface Hydrology Program for their purchase of the Oklahoma Mesonet soil moisture data for NLDAS investigators, Illinois State Water Survey for providing soil moisture observations in Illinois State,

References (67)

  • A. Betts et al.

    Assessment of the land surface and boundary layer models in two operational versions of the NCEP Eta model using FIFE data

    Mon. Weather Rev.

    (1997)
  • F.V. Brock et al.

    The Oklahoma Mesonet: a technical overview

    J. Atmos. Ocean. Technol.

    (1995)
  • Burnash, R.J.C., Ferral, R.L., McGuire, R.A., 1973. A Generalized Streamflow Simulation System: Conceptual Models for...
  • T.H. Chen et al.

    Cabauw experimental results from the project for intercomparison of land-surface parameterization schemes

    J. Clim.

    (1997)
  • R.B. Clapp et al.

    Empirical equations for some soil hydraulic properties

    Water Resour. Res.

    (1978)
  • W.T. Crow et al.

    Multi-scale dynamics of soil moisture variability observed during SGP97

    Geophys. Res. Lett.

    (1999)
  • W. Crow et al.

    Upscaling sparse ground-based soil moisture observations for the validation of satellite surface soil moisture products

    Rev. Geophys.

    (2011)
  • T. Delworth et al.

    The influences of potential evaporation on the variabilities of simulated soil wetness and climate

    J. Clim.

    (1988)
  • I. Dharssi et al.

    Operational assimilation of ASCAT surface soil wetness at the Met Office

    Hydrol. Earth Syst. Sci. Discuss.

    (2011)
  • P. Dirmeyer

    Using a global soil wetness data set to improve seasonal climate simulation

    J. Clim.

    (2000)
  • M. Drusch et al.

    Assimilation of screen-level variables in ECMWF’s integrated forecast system: a study on the impact on the forecast quality and analyzed soil moisture

    Month. Weather Rev.

    (2007)
  • M. Drusch et al.

    Towards a Kalman Filter based soil moisture analysis system for the operational ECMWF Integrated Forecast System

    Geophys. Res. Lett.

    (2009)
  • M.B. Ek et al.

    Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model

    J. Geophys. Res.

    (2003)
  • J. Entin et al.

    Evaluation of Global Soil Wetness Project soil moisture simulations

    J. Meteorol. Soc. Jpn.

    (1999)
  • J. Entin et al.

    Temporal and spatial scales of observed soil moisture variations in the extratropics

    J. Geophys. Res.

    (2000)
  • Y. Fan et al.

    1948–98 U.S. Hydrological reanalysis by the Noah land data assimilation system

    J. Clim.

    (2006)
  • H. Gao et al.

    Copula derived observation operators for assimilating TMI and AMSR-E soil moisture into land surface models

    J. Hydrometeorol.

    (2007)
  • Z. Guo et al.

    Improving the quality of simulated soil moisture with a multi-model ensemble approach

    Q. J. R. Meteorol. Soc.

    (2007)
  • Z. Guo et al.

    Land surface impacts on subseasonal and seasonal predictability

    Geophys. Res. Lett.

    (2011)
  • C.R. Hain et al.

    Retrieval of an available water-based soil moisture proxy from thermal infrared remote sensing. Part I: Methodology and validation

    J. Hydrometeorol.

    (2009)
  • S.E. Hollinger et al.

    A soil moisture climatology of Illinois

    J. Clim.

    (1994)
  • B.G. Illston et al.

    Mesoscale monitoring of soil moisture across a statewide network

    J. Atmos. Ocean. Technol.

    (2007)
  • M. Kanamitsu et al.

    NCEP-DOE AMIP-II reanalysis (R-2)

    Bull. Am. Meteorol. Soc.

    (2002)
  • Cited by (156)

    View all citing articles on Scopus
    View full text